Jules_nerd wrote on 2026-08-12, 20:24:

Well, there really isn't much to go wrong on the PS2 mouse port in this design, so long as that input on the M6117 is working...
Did you get a chance to check it out?

Checked the connections from the PS/2 port to the M6117 which seem okay. Also checked a couple of resistors and capacitors and they check out okay as well.

I'm out of ideas at the moment. So going to reassemble the laptop again. Can only hope that the flux I removed was conductive and that it magically is solved. 😜

Jules_nerd wrote on 2026-08-12, 08:23:

Incidentally, I assume that on the BOOK you can see BIOS config options for LPT2? In the soundcard manual's description of the COVOX it mentions that a real LPT2 hardware must exist if your COVOX software looks for a parallel port. If the BIOS allows you to set a different address for LPT2 it may 'break' covox software that does these checks; Whilst BIOS may be fine setting the 'real' LPT2 address in the M5113, the COVOX has its own, separate address decoding is fixed in hardware to the default address.

I forgot to reply to this. No, LPT2 settings do not show at all in the BIOS setup -- same as the Pocket386.
I am much less familiar with parallel. I don't even know of a good program to play with it in DOS --- I use Kermit for serial.

By the way, on the topic of the touchpad and its frequent mis-detections, it seems that both OS/2 and Linux have no problem: they consistently detect it on each boot with no issue.
So it must be something related to the initial handshake that MS mouse drivers (specially Win95) don't like, rather than a HW fault. (Albeit since we can't change the FW....).

Mine has arrived now..
Interesting, not only is there no setup for an LPT2, but diagnostics software seems to show that there isn't one! 😁 it also shows no IRQ assigned to LPT1 on mine atm.
So im not sure why they have said what they have in the sound-card manual. The COVOX works fine you may need to tell any software that its at the default address for LPT2. Though if the software you use for covox actually checks for a real LPT2 at that address it should fail! A real parallel port (even the original unidirectional ones) have input registers that can read the outputs, so it can read back any data you output, this is used to check for a port amongst other things. These input registers aren't present on the Book's COVOX circuits...

Well, congrats and good hunting (for HW bugs!) 😀

Yeah, about the Covox personally I don't think it's that useful .
It may provide a couple DOS games with voice, which is nice of course.
And those games likely won't care too much about the "quality" of the LPT port? Haven't tried myself.

This is likely to be only a problem if you try to use it from a real OS that may probe the ports in some way.

But e.g. for using it for PCM audio output for a real OS ... it just burns way too many CPU cycles. And you could already output PCM through the Adlib if you were willing to burn lots of CPU cycles.
i.e. to hear the usual win31/win95 jingles, even with the Covox present, I think it may be more efficient to continue (ab)using the adlib with the adlibw driver or similar ...

I must admit I liked the cute form factor of the pocket, but the book being considerably larger & including a larger keyboard does make it a more versatile, usable machine.... its a much easier keyboard to use
Im not sure about the resolution of the screen either - there are some very thin horizontal lines that flash up when its changing video modes, they are maybe thinner than Id have thought for 800x600?? its hard to judge...

It looks good though, and if you were weighing up getting one of these vs a real vintage 386laptop, the screen is far, far better than Id expect from a 386 era machine - in fact its better than the 800x600 screen on my pakard Bell easynote, which is a PII era machine...
The OSD is much better than the pocket, more options & no weird buggy looking characters flashing up on the screen!

That fan noise is rather annoying! Whilst the circuit just switches the power to the fan on/off (about 6 to 8 volts on battery, 12V when powered by the adapter), it seems it has been designed to do PWM speed control - you can hear it takes a while to speed up when first switched on with the 12V plugged in. The OSD gives a voltage reading for an NTC, I presume this is the thermistor RT1 which is placed on the PCB near the power supply regulators to measure temperature.
Soo it seems the hardware has been designed to give a temperature related fan speed, but this doesnt seem to have been implemented, or at least not well!
Once it has got up to speed the fan seems to stayt at the same speed regardless of the NTC reading!!!
Also the readings dont make 100% sense; its supposed t be a 10k Thermistor (10k at 25 degree C) in the lower half of a potential divider with R152 (10k) from +5V, so @ 25 degrees C you should get half the supply, about 2.5V ish, yet mine shows much higher at switch on... Its as if R152 is actually higher than the 10k on the schematic? The voltage does drop as it warms up though & drops further when you switch off the fan.... so that seems to work to some extent.
But is still doenst seem to change the fan speed...
There isnt much hardware here for us to change, so we are again at the mercy of the firmware in the 8051.....
Shame I don't know what coefficient the NTC is , or we could have a stab at calculating the temperature (wouldn't it be nice if the 8051 software did this for us?)

javispedro1 wrote on 2026-08-17, 19:48:

Yeah, about the Covox personally I don't think it's that useful . It may provide a couple DOS games with voice, which is nice of […]
Show full quote

Yeah, about the Covox personally I don't think it's that useful .
It may provide a couple DOS games with voice, which is nice of course.

But e.g. for using it for PCM audio output for a real OS ... it just burns way too many CPU cycles. And you could already output PCM through the Adlib if you were willing to burn lots of CPU cycles.
i.e. to hear the usual win31/win95 jingles, even with the Covox present, I think it may be more efficient to continue (ab)using the adlib with the adlibw driver or similar ...

I suppose if one does have problems with software not detecting a parallel port, you could always try to set LPT1's address to the default LPT2 address (278) and that should work OK. The COVOX registers just read whatever is sent to that address, it shouldnt care that they are being read by the parallel port as well...
But yes, Im with you, COVOX isnt much more than a novelty really... I guess they did it because its easy/cheap & better than nothing..
But Im still considering adding a proper sounblaster or similar, but I haven't opened up the case yet to see where it could be fitted. Im not a fan of putting SMD parts on both sides of a PCB, but I guess it could help fit it all on that soundcard PCB.

I was wrong about the coprocessor. Ive got an IIT in mine thats only rated for 33MHz.
Im not a fan of overclocking really, but its what Ive got.. for now at least.
Seems to work OK (but Ive not tried tests that actually verify the accuracy of the results, apparently this can be an issue when overclocking them), & only increases the supply current by around 100mA or so, so about 500mW.
Even when I give it a 'Damn good thrashing' running SPICE simulations it surface temp only gets to 20 degrees C above ambient... no surprises, its a big package for 500mW....
So it may well be fine with a cover over it!
YMMV with other coprocessors of course, even with other examples of the same type...

Jules_nerd wrote on 2026-08-17, 20:15:

That fan noise is rather annoying! Whilst the circuit just switches the power to the fan on/off (about 6 to 8 volts on battery, 12V when powered by the adapter), it seems it has been designed to do PWM speed control - you can hear it takes a while to speed up when first switched on with the 12V plugged in. The OSD gives a voltage reading for an NTC, I presume this is the thermistor RT1 which is placed on the PCB near the power supply regulators to measure temperature.

Well that explains why it becomes incredibly annoying when plugged in.
Thanks for NTC heads-up ! I had no idea what it was for. Will monitor it in the future.

Jules_nerd wrote on 2026-08-17, 20:15:

The OSD is much better than the pocket, more options & no weird buggy looking characters flashing up on the screen!

While I don't know what the buggy characters were 😜 , for me it still has the same bugs, such as the eventual desynchronization of whether the 8051 forwards keys to the OSD or not . One way to reliably trigger it is the auto-sync feature, which hides the OSD but the 8051 doesn't notice so it will keep eating the keys. Also, I am yet to figure out the logic for the navigation keys...

I really, really wish the 8051 could be programmed...

Jules_nerd wrote on 2026-08-17, 20:22:

I suppose if one does have problems with software not detecting a parallel port, you could always try to set LPT1's address to the default LPT2 address (278) and that should work OK. The COVOX registers just read whatever is sent to that address, it shouldnt care that they are being read by the parallel port as well...

If anyone fancies a simple soldering job, you can make this covox look like the FTL 'adapter'; in the BIOS set the parallel port to 278 & solder yourself a 25 pin male D type with a link between pins 12 & 17 (or use a 220 Ohm resistor if your worried about damaging anything)... Dungeon master will then detect it and give you the option to select FTL for sound.

javispedro1 wrote on 2026-08-17, 21:36:

I really, really wish the 8051 could be programmed...

Well I don't know what protection options have been set in the STC8G2K64S4 (or STC8G2K32S2 in the pocket)... But even if you could read the code, would anyone be up for reverse engineering?
Putting new code in is always possible, even if you have to solder a new device in... unfortunately I don't even see a PCB footprint for an in circuit programmer, but Im sure something could be hacked on for dev work.....
But either way, whether nor not you can read the code out, not being a programmer I assume this is a pretty big task?....
I presume the only way it might be viable is if they released the source?

javispedro1 wrote on 2026-08-17, 21:36:

Well that explains why it becomes incredibly annoying when plugged in.
Thanks for NTC heads-up ! I had no idea what it was for. Will monitor it in the future.

Well when starting from cold (room @22 degrees C ) mine showed an NTC reading of about 3700 mV with a +5V of 4.90V (this would imply the 10k pull up has been changed to 2k7, if the NTC is still a 10k one)
With the fan off this seemed to drop down to 1325mV (I got bored waiting, it might get lower...)
Fan on running form battery seems to plateau at 2250mV
With fan on, running from the 12V adapter, 2500mV

The fan speed does seem to vary with this range of readings, it just maxes out at the equilibrium point and sits there....

So It seems that even running at the slower speed from the battery it is flowing enough air to pretty much do the job, maybe some mod limiting the max voltage might be in order...
one option might be to put the fan in the other way around, at the moment it sucks air in from the side of the book, blows it straight over the main PSU regulator & over the NTC, then the M6117, also the graphics chip above them.
This gives the best cooling of the hot parts, but does make the rest of the product hotter than if the fan was the other way around. If you do this the NTC is really sampling cool air to a degree, though the copper layers in the PCB will no doubt help it 'sample' the temp of the main regulator....
From the pics I can see it looks as though the route for air from the vents on the bottom does seem rather long, it looks as though the PCB goes pretty much up to the inside of the case most of the way around, apart from at the front near the network connector??? maybe some more vents might help.. perhaps in the metal plates around the connectors...

ATM I'm loath to take it apart; Ive yet to see anything so wrong it needs fixing (and can be fixed easily) other than the RTC clock on mine has 'jumped' by 20 mins or so on a power cycle - indicating that there is still a lurking power off/on reset/time/BIOS issue, though this seems much, much less frequent/serious than on the pocket...

Until I get the lid off & look at the parts fitted (and even then Im not sure we will know the NTC fitted) the rest is pretty much guesswork... Certainly if I do take it in bits Ill take lots of pics!

Jules_nerd wrote on 2026-08-18, 19:14:

With fan on, running from the 12V adapter, 2500mV

After lengthy session mine was showing 900mV, even with the fan on....
Fan is practically on the same level of noise for me, no matter the value of NTD. Only when unplugged I can hear some variation.

Jules_nerd wrote on 2026-08-18, 19:14:

Well I don't know what protection options have been set in the STC8G2K64S4 (or STC8G2K32S2 in the pocket)... But even if you could read the code, would anyone be up for reverse engineering?
Putting new code in is always possible, even if you have to solder a new device in... unfortunately I don't even see a PCB footprint for an in circuit programmer, but Im sure something could be hacked on for dev work.....
But either way, whether nor not you can read the code out, not being a programmer I assume this is a pretty big task?....

Well it is an interesting question because I have zero experience with it, but something like updating the keyboard layout should be trivial at least (or changing the fan algorithm?) . But I am not sure we can read the existing program at all, and if this requires significant resoldering I'd probably give up in advance, though 🙁

One of my motivations for the book386 is so that I can have a 386-486-Pentium machine to test software, that I can easily replace -- so I do not feel as guilty as when I inevitably break a real 20-30 year old relic due to overuse, fragile plastics, or whatever.
But with the book386 at $200 ... well, I'm still hesitating to open it 😀

In other topics:
I got the floppy cable from the same seller, and it works with a Gotek. I also can boot from it.

So far I'm still spending my Book386 time trying to fill in a 16GB CF with as many OSes I can think of. Do not recommend, as for almost any OS you want to run on a 386, you'll need a DDO (ontrac / ez-disk) for the >8GB. Even Linux/LILO seems to have only gotten serious LBA support starting with Debian 2.2 Potato, in 2000...

javispedro1 wrote on 2026-08-18, 20:23:

In other topics:
I got the floppy cable from the same seller, and it works with a Gotek. I also can boot from it.

Good to know!

There is now also a complete floppy disk drive available for the BOOK386: https://www.tindie.com/products/cycle/laptop- … ppy-disk-drive/ for 30 Euro. This includes the floppy adapter, drive , cable etc.

Those floppy options are not very usable for me as I prepare the CF card in VMware Workstation Pro and if needed with a DOS software floppy emulator. When I receive my ISA expansion (and have soldered it) I will test a multi IO card and try to get a floppy drive working with that. Just for giggles. Main use will be sound cards etc. Curious if the 3rd ISA slot is working as well if soldered on (maybe needs needs external power?).

Yoghoo wrote on 2026-08-19, 11:53:

Curious if the 3rd ISA slot is working as well if soldered on (maybe needs needs external power?).

I just soldered mine up, I haven't fitted a third connector, but I did 'bell out' all the connections before I soldered it up, the third connector footprint is identical to the other two, wired the same, the connector fits the same etc etc, so I can see no reason why it wouldn't work. Only difference is the extra labeling on it, as I said I presume to make it easy for people to use to add their own ad-hoc circuits. I think all three are spaced at the standard 20.3mm
Personally Id have designed the pads on the edge card footprint differently - on the solder side is the same small annulus you get on the component side, which is a bit small, it would be easier to hand solder with a bigger ob-round pad on the solder-side (maybe I'm old fashioned!).
Not a big deal for those used to soldering PTH boards, but a beginner might find it a little awkward.

All the power supply rails are wired straight from the book connectors through to the edge card connectors & the 'berg' footprint. I think this is better than trying to use external supplies, this way you know they come up at the right time compared with RESET...
The BOOKs PSU looks considerably beefed up compared to the pocket...
The only SMD components (EDIT, already fitted to the PCB) are supply decoupling caps & an LED with series resistor on the +5V

Received the ISA extension kit yesterday. Was already pre-solderend so that was a nice surprise. An ALS-100 sound card was detected but OPL3 came from the laptop speaker (as suspected). As there is no line-in you need to use a audio Y-cable to redirect both outputs to external speakers. Or setup the sound card differently but that would still need external speakers and would probably hurt game comparability.

Also played with a controller card. Extra parallel port is working but only LPT1 and LPT2. If I set it up as LPT3 it becomes LPT1 and the laptop LPT1 becomes LPT2. COVOX was still working with both LPT ports defined. Also it seems the controller card will 'overwrite' the game port and IDE ports if enabled. Didn't try to get an external HDD to work btw. It just couldn't boot and asked for a floppy. Couldn't get a floppy drive to work during my quick tests. But I suspect that has something to do with the controller card.

Forgot to mention that my PS/2 mouse port is still not working so will probably disassemble it in the future again. Does somebody know if the BIOS is available somewhere? Forgot to make a backup when I disassembled it.

Yoghoo wrote on 2026-08-23, 09:05:

Forgot to mention that my PS/2 mouse port is still not working so will probably disassemble it in the future again. Does somebody know if the BIOS is available somewhere? Forgot to make a backup when I disassembled it.

In case there's a glitch in some common source code, at BIOS or KBC level, and because the same ppl implemented the hardware, a very specific driver recommendation.. Re: Pocket 8086 - All the information you need.

BitWrangler wrote on 2026-08-23, 12:45:

Yoghoo wrote on 2026-08-23, 09:05:

Forgot to mention that my PS/2 mouse port is still not working so will probably disassemble it in the future again. Does somebody know if the BIOS is available somewhere? Forgot to make a backup when I disassembled it.

In case there's a glitch in some common source code, at BIOS or KBC level, and because the same ppl implemented the hardware, a very specific driver recommendation.. Re: Pocket 8086 - All the information you need.

Thanks. But unfortunately it's a hardware issue. I tried around 6 different mouse drivers from different vendors including that version of ctmouse (and the latest as well). All did not work.

Jules_nerd wrote on 2026-08-17, 20:22:

I suppose if one does have problems with software not detecting a parallel port, you could always try to set LPT1's address to the default LPT2 address (278) and that should work OK. The COVOX registers just read whatever is sent to that address, it shouldnt care that they are being read by the parallel port as well...

As a more risky alternative, one can write to the BIOS data area that LPT2 is at 278$, and this may be enough to fool a lot of software that doesn't bother to test what's on that IO port. You can keep LPT1 working as normal.
You can do this with plain debug.exe, open a debug session and write

1e 40:0a 78 02

<enter> then q.
For me this enough to fool MS Diagnostics...

Yoghoo wrote on 2026-08-23, 12:58:

Thanks. But unfortunately it's a hardware issue. I tried around 6 different mouse drivers from different vendors including that version of ctmouse (and the latest as well). All did not work.

Did you check the PS2 connector itself? sounds like you checked pretty much everything else...
Check for shorts as well as continuity?